Real Estate Price Prediction using Machine Learning and Data Analytics

Abstract

In this paper we presents a complete model to predict Real Estate prices with high efficiency through Machine Learning (ML) and Data Analytics approach. The model data is based on the large-scale real estate property data containing structural, locational and environmental elements to become the basics of price variation predictors. We pre-processed, feature engineered and analysed 50,000 Land Registry compliant datasets using a variety of machine learning models - Linear Regression, Random Forest, XGBoost and ANN. Random Forest had the best predective capacity with a Mean Absolute Error (MAE) of 2.63 lakhs and R² value of 0.8732 which indicates a high generalisation ability and is very strong. This paper suggests that the challenge of real estate pricing can be addressed by using data-driven analytics, ensemble learning and intelligent feature engineering. The results also indicate the effectiveness of the advanced ML to both the real-world real estate valuation and market forecasting, in addition decision making in property investment.